Search

CN-119991479-B - Underwater image enhancement method, system and equipment combined with attenuation channel correction

CN119991479BCN 119991479 BCN119991479 BCN 119991479BCN-119991479-B

Abstract

The invention relates to an underwater image enhancement method, an underwater image enhancement system and underwater image enhancement equipment combined with attenuation channel correction, wherein the method comprises the following steps of carrying out color correction on an image by adopting a white balance technology of attenuation channel self-adaptive correction, marking the image as a first enhancement image I 1 , carrying out denoising treatment on the first enhancement image I 1 by utilizing a rapid non-local mean value algorithm based on local blocks, marking the image as a second enhancement image I 2 , improving the contrast of the first enhancement image I 1 by adopting a self-adaptive stretching method based on local characteristics of a histogram, marking the image as a third enhancement image I 3 , and combining the complementary advantages of the second enhancement image I 2 and the third enhancement image I 3 by utilizing a bilateral weight fusion strategy to obtain a final enhancement image. The system and apparatus are for performing the above method. According to the invention, qualitative and quantitative experimental results on different underwater image reference data sets show that the contrast of the image can be uniformly improved and the details and textures of the image can be enhanced while correcting the color of the image.

Inventors

  • XIANG DAN
  • HE DENGYU
  • LIANG NAIYAO
  • LIU YUKUN
  • LIU FANG

Assignees

  • 广州航海学院

Dates

Publication Date
20260512
Application Date
20250114

Claims (10)

  1. 1. A method of underwater image enhancement incorporating attenuation channel correction, comprising the steps of: s01, acquiring an original image, performing white balance processing on the original image to perform color correction, and marking the image after the color correction as a first enhancement chart I 1 ; S02, processing the first enhancement map I 1 through a non-local mean denoising algorithm to obtain a denoised image, marking the denoised image as a second enhancement map I 2 , enhancing the contrast of the first enhancement map I 1 through an adaptive stretching method, and marking the image with enhanced contrast as a third enhancement map I 3 ; s03, calculating to obtain a saliency weight map, a contrast weight map and a gradient weight map related to the second enhancement map I 2 , and carrying out weight normalization processing to obtain a first normalized weight map; S04, decomposing the second enhancement map I 2 and the third enhancement map I 3 into a Laplacian pyramid structure, decomposing the first normalized weight map and the second normalized weight map into a Gaussian pyramid structure, and carrying out bilateral weight fusion on the Laplacian pyramid structure and the Gaussian pyramid structure obtained by decomposition to obtain the enhancement image.
  2. 2. The underwater image enhancement method in combination with attenuation channel correction as claimed in claim 1, wherein the step S01 comprises the steps of: s011, calculating index parameters of RGB channels of the original image according to the following formula, wherein the index parameters comprise information entropy Length of interval Average degree of difference : , , , Wherein, the , Representation of The entropy of the information of the channel, Representation of Channel No Probability at the individual pixel level; Representation of The number of pixels of the channel is greater than or equal to Is set for the length of the interval of (c), Represent the first The number of pixels at the level of a pixel, And Respectively indicate when Time of day Is set to be the minimum value and the maximum value of (c), Representing the total number of pixel levels, Representing a pixel count threshold; Representation of Average degree of difference of channels; s012, giving weight to the index parameters and constructing compensation parameters of each channel in the original image : , Wherein, the Representation of The compensation parameters of the channels are set to be, 、 And Respectively represent information entropy Length of interval Average degree of difference Weight coefficient of (2); S013, according to the compensation parameters The RGB channels are ordered according to the numerical values of the channels, and the RGB channels are sequentially recorded as a first channel according to the order from the big value to the small value A second channel And a third channel According to the compensation parameters of each channel Assigning values to the channels, and constructing a compensation factor and a compensation formula according to the sequencing and the assignment results: , , Wherein, the And Representing a first compensation factor and a second compensation factor respectively, And Respectively representing the compensated second channel and the third channel; and S014, processing the compensated image through a gray world algorithm to obtain the first enhancement map I 1 after color correction.
  3. 3. The method for enhancing an underwater image combined with attenuation channel correction according to claim 1, wherein in step S02, the first enhancement map I 1 is processed by a non-local mean denoising algorithm, and the obtained denoised image is recorded as a second enhancement map I 2 , which specifically comprises the following steps: s021a, reading the first enhancement map I 1 , defining the length of the local block as Width of ; S022a, initializing an image with the same size as the first enhancement chart I 1 , and recording the image as a blank image; s023a, calculating similarity weight in the local block according to the following formula: , Wherein, the Representing target pixels Similar to a pixel The weight of the similarity between the two, Representing target pixels in a local block Is used for the display of the display panel, Representing similar pixels in a local block Is used for the display of the display panel, Weight parameters for controlling the degree of influence of the similarity weights; s024a, calculating the weighted average value of similar pixels in the local block according to the similarity weight to obtain a denoising result : , S025a, reading R, G, B three channels in parallel, and calculating the denoising result on each channel Denoising the result of each channel All are stored in the blank image, and the second enhancement map I 2 is obtained.
  4. 4. The method for enhancing an underwater image in combination with attenuation channel correction according to claim 3, wherein in step S02, the contrast of the first enhancement map I 1 is enhanced by an adaptive stretching method, and the image after the contrast enhancement is obtained is denoted as a third enhancement map I 3 , specifically comprising the following steps: S021b, reading the RGB channel histogram of the first enhancement chart I 1 , and calculating the exposure value and the exposure threshold value of the obtained RGB channel histogram: , , Wherein, the The exposure value is indicated as such, Indicating exposure threshold, using Dividing the RGB channel histogram into two sub-histograms, namely a first sub-histogram and a second sub-histogram, wherein the distribution range of the first sub-histogram is within The distribution range of the second sub-histogram is within the low exposure region of Is used for the exposure of the substrate to the light, Represent the first The number of pixels at the level of a pixel, Representing a pixel count threshold; s022b, clipping the first sub-histogram and the second sub-histogram by adopting the median value of the RGB channel histogram as a clipping threshold value, clipping the part with the total number of pixels larger than the clipping threshold value, keeping the part with the total number of pixels smaller than or equal to the clipping threshold value unchanged, redistributing the total number of clipped pixels into the whole interval of the sub-histogram, and reserving image information, wherein the calculation formula of the clipping process is as follows: , Wherein, the Representing the sub-histogram after clipping, Representing the total number of pixels that are clipped, The median value of the sub-histogram is represented, Representing bin lengths of the sub-histograms; s023b, constructing a stretching factor according to the interval length ratio of the sub-histograms before and after stretching, and obtaining the stretched sub-histogram according to a rule of proportional stretching: , , Wherein, the And The stretching factors of the first sub-histogram and the second sub-histogram are represented respectively, And Respectively representing the sub-histograms after stretching, the corresponding stretching ranges are respectively And ; S024b, combining the two sub-histograms after the stretching treatment to obtain an image with enhanced contrast, namely the third enhancement chart I 3 .
  5. 5. The method of underwater image enhancement in combination with attenuation channel correction as claimed in claim 4, wherein the step S03 comprises the steps of: s031, calculating the significance weights of the second enhancement map I 2 and the third enhancement map I 3 according to the following formula: , Wherein, the The position of the object is indicated and, Represent the first The saliency weight of the individual input images, 、 And Respectively represent the first Luminance channel, red-green channel and yellow-blue channel in the CIELAB color space of the individual input images, 、 And Respectively the first Average values corresponding to the brightness channel, the red-green channel and the yellow-blue channel of the input images; Gradient weights of the second enhancement map I 2 and the third enhancement map I 3 are calculated as follows: , Wherein, the The position of the object is indicated and, Represent the first The gradient weights of the images of the individual input images, And Respectively represent the first Gradients of images of the input images in the horizontal direction and the vertical direction; The noise weight of the third enhancement map I 3 is calculated as follows: , , Wherein, the The position of the object is indicated and, For the weight of the noise to be given, Representing the position At the local noise level of the location(s), And The height and width of the partial window respectively, The size of the half window is indicated, Representing pixels in a local window Is used for the display of the display panel, Representing the mean of the local window; S032, dividing the sum of the weights of each enhancement map by the sum of the weights of all enhancement maps to respectively normalize each enhancement map, wherein the calculation formula is as follows: , Wherein, the Represent the first Normalized weight map of the individual input images, Represent the first The weighted sum of the individual input images, Representing regularization terms.
  6. 6. The method for enhancing an underwater image by combining attenuation channel correction according to claim 5, wherein in step S04, bilateral weight fusion is performed on the laplacian pyramid structure and the gaussian pyramid structure obtained by decomposition, so as to obtain an enhanced image, which specifically comprises: , Wherein, the An enhanced image is represented and is displayed, Representing the post-decomposition Gaussian pyramid First of the input images A layer weight map is provided for each layer, Representing the decomposed Laplacian pyramid First of the input images The image of the layer is a layer image, Indicating the number of layers decomposed.
  7. 7. The method of underwater image enhancement in combination with attenuation channel correction as claimed in claim 6, wherein the number of layers decomposed 。
  8. 8. An underwater image enhancement system incorporating attenuation channel correction, comprising: The color correction module is used for acquiring an original image, performing white balance processing on the original image to perform color correction, and recording the image after the color correction as a first enhancement chart I 1 ; The enhancement module is used for processing the first enhancement map I 1 through a non-local mean denoising algorithm, obtaining a denoised image and marking the denoised image as a second enhancement map I 2 , enhancing the contrast of the first enhancement map I 1 through an adaptive stretching method, and obtaining a contrast-enhanced image and marking the contrast-enhanced image as a third enhancement map I 3 ; The normalization processing module is used for calculating and obtaining a saliency weight map, a contrast weight map and a gradient weight map of the second enhancement map I 2 and carrying out weight normalization processing to obtain a first normalization weight map; And the fusion module is used for decomposing the second enhancement map I 2 and the third enhancement map I 3 into a Laplacian pyramid structure, decomposing the first normalized weight map and the second normalized weight map into a Gaussian pyramid structure, and carrying out bilateral weight fusion on the Laplacian pyramid structure and the Gaussian pyramid structure obtained by decomposition to obtain the enhancement image.
  9. 9. A computer device comprising a processor and a memory in signal connection, characterized in that the memory has stored therein at least one instruction or at least one program, which when loaded by the processor performs the underwater image enhancement method in combination with attenuation channel correction as claimed in any of claims 1-7.
  10. 10. A computer readable storage medium having stored thereon at least one instruction or at least one program, wherein the at least one instruction or the at least one program when loaded by a processor performs the underwater image enhancement method in combination with attenuation channel correction as claimed in any of claims 1-7.

Description

Underwater image enhancement method, system and equipment combined with attenuation channel correction Technical Field The invention relates to the technical field of underwater image processing, in particular to an underwater image enhancement method, an underwater image enhancement system and underwater image enhancement equipment combined with attenuation channel correction. Background In recent years, underwater imaging plays a vital role in underwater vehicles, deep sea exploration, underwater target recognition, and the like. The underwater image is an important information carrier for acquiring ocean resource information, but due to the complex and diversified underwater imaging environments and the wavelength-dependent light absorption and scattering, the problems of color deviation, low contrast and blurred details of the underwater image are caused. In particular, as light propagates in aqueous media, it undergoes varying degrees of attenuation, in general blue and green light propagate relatively well in water, while red and other wavelengths of light are more readily absorbed. Thus, the light reflected by the underwater object contains relatively more of the cyan component, resulting in the underwater image exhibiting a characteristic blue-green appearance. Meanwhile, particles and suspended matters in the aquatic medium can cause light scattering phenomenon, so that light rays are scattered in multiple directions, and finally the contrast of an underwater image is reduced, and the overall quality of the image is damaged. In the prior art, how to enhance the underwater image quality mainly starts from improvement based on four directions of hardware, recovery, deep learning and enhancement, and the difficulty and the cost of the improvement based on hardware are high due to unstable aquatic environment and high equipment cost. The main limitations of restoration-based methods are the difficulty in adapting to complex underwater environments and the severe reliance on a priori knowledge and parameter selection. The deep learning-based method has the problems of limited generalization capability, complex parameter setting and high calculation complexity. The enhancement-based improvement has advantages of easy implementation and high efficiency compared to the aforementioned three methods, but it is difficult to comprehensively solve various problems in the underwater image by the enhancement-based method, resulting in the situation that over enhancement or under enhancement is liable to occur in the enhanced image. As disclosed in the chinese patent application document of application No. 202310657636.2, a multi-scale fusion underwater image enhancement method combining with adaptive gamma correction is disclosed, in which color correction and contrast enhancement are performed on an image through white balance and adaptive gamma correction, and then a final enhanced image is obtained by using a fusion algorithm. Although the above method can perform color correction and contrast enhancement on the original image, noise existing in the original image is not considered, so that details of the finally enhanced image are not emphasized because of the noise. To sum up. In the prior art, the problem of image noise is not fully considered in the method for enhancing the underwater image, so that the enhancement effect is general. Disclosure of Invention In order to solve the problems of the prior art, the invention aims to provide an underwater image enhancement method, an underwater image enhancement system and underwater image enhancement equipment combined with attenuation channel correction. According to the invention, the original image is subjected to color correction by a white balance technology, the color cast problem is solved, the color corrected image is denoised by adopting a rapid non-local mean value algorithm based on local blocks, meanwhile, the contrast enhancement is performed by adopting a self-adaptive stretching method based on the local characteristics of a histogram, and finally, a bilateral weight fusion strategy is introduced to combine the complementary advantages of the denoised image and the contrast enhanced image, so that the underwater image is effectively enhanced, and meanwhile, the details and the edge textures of the image can be highlighted. The invention relates to an underwater image enhancement method combined with attenuation channel correction, which comprises the following steps: s01, acquiring an original image, performing white balance processing on the original image to perform color correction, and marking the image after the color correction as a first enhancement chart I 1; S02, processing the first enhancement map I 1 through a non-local mean denoising algorithm to obtain a denoised image, marking the denoised image as a second enhancement map I 2, enhancing the contrast of the first enhancement map I 1 through an adaptive stretching method, and marking